\ub525\ub7ec\ub2dd(Deep Learning)\uc740 \uc778\uacf5\uc2e0\uacbd\ub9dd\uc744 \uae30\ubc18\uc73c\ub85c \ud55c \uba38\uc2e0\ub7ec\ub2dd\uc758 \ud55c \ubd84\uc57c\ub85c, \ub370\uc774\ud130\uc5d0\uc11c \ud328\ud134\uc744 \ud559\uc2b5\ud558\uc5ec \uc608\uce21\uc744 \uc218\ud589\ud558\ub294 \uae30\uc220\uc785\ub2c8\ub2e4. \ub525\ub7ec\ub2dd\uc740 \uc0ac\ub78c\uc758 \ub1cc \uad6c\uc870\uc5d0\uc11c \uc601\uac10\uc744 \ubc1b\uc544 \ub2e4\uce35 \uc2e0\uacbd\ub9dd\uc744 \uc0ac\uc6a9\ud574 \uc785\ub825 \ub370\uc774\ud130\uc5d0\uc11c \uc801\uc808\ud55c \ube44\uc120\ud615 \ubcc0\ud658\uc744 \ud1b5\ud574 \uc774\ud574\ud558\uace0 \ud559\uc2b5\ud558\ub294 \uacfc\uc815\uc744 \uac70\uce69\ub2c8\ub2e4. \uc774\ub294 \uc774\ubbf8\uc9c0 \uc778\uc2dd, \uc790\uc5f0\uc5b4 \ucc98\ub9ac, \uc74c\uc131 \uc778\uc2dd \ub4f1 \ub2e4\uc591\ud55c \ubd84\uc57c\uc5d0\uc11c \ud65c\uc6a9\ub429\ub2c8\ub2e4.<\/p>\n<\/section>\n \ud30c\uc774\ud1a0\uce58(PyTorch)\ub294 \ud398\uc774\uc2a4\ubd81 \uc778\uacf5\uc9c0\ub2a5 \uc5f0\uad6c\ud300\uc5d0\uc11c \uac1c\ubc1c\ud55c \uc624\ud508\uc18c\uc2a4 \uba38\uc2e0\ub7ec\ub2dd \ub77c\uc774\ube0c\ub7ec\ub9ac\uc785\ub2c8\ub2e4. \ud30c\uc774\ud1a0\uce58\ub97c \uc0ac\uc6a9\ud558\uba74 \ub525\ub7ec\ub2dd \ubaa8\ub378\uc744 \uad6c\uc131\ud558\uace0 \ud6c8\ub828\ud558\ub294 \uacfc\uc815\uc744 \uc27d\uac8c \uad6c\ud604\ud560 \uc218 \uc788\uc73c\uba70, \ud2b9\ud788 \ub3d9\uc801 \uacc4\uc0b0 \uadf8\ub798\ud504\ub97c \uc9c0\uc6d0\ud558\uc5ec \ub354 \uc9c1\uad00\uc801\uc778 \ubaa8\ub378 \uad6c\ucd95\uc774 \uac00\ub2a5\ud569\ub2c8\ub2e4. \ud30c\uc774\ud1a0\uce58\ub294 \uc8fc\ub85c \ud30c\uc774\uc36c\uc73c\ub85c \uc791\uc131\ub418\uc5c8\uc73c\uba70, GPU\ub97c \ud65c\uc6a9\ud55c \uace0\uc18d \uc5f0\uc0b0\uc774 \uac00\ub2a5\ud569\ub2c8\ub2e4.<\/p>\n<\/section>\n \uc778\uacf5 \uc2e0\uacbd\ub9dd\uc740 \uc0dd\ubb3c\ud559\uc801 \uc2e0\uacbd\ub9dd\uc758 \uad6c\uc870\uc5d0\uc11c \uc601\uac10\uc744 \ubc1b\uc544 \uac1c\ubc1c\ub41c \ubaa8\ub378\uc785\ub2c8\ub2e4. \uc5ec\ub7ec \uce35\uc73c\ub85c \uad6c\uc131\ub418\uc5b4 \uc788\uc73c\uba70, \uac01 \uce35\uc740 \uc785\ub825 \uc2e0\ud638\ub97c \ucc98\ub9ac\ud558\uc5ec \ub2e4\uc74c \uce35\uc73c\ub85c \uc804\ub2ec\ud569\ub2c8\ub2e4.<\/p>\n<\/li>\n \uc190\uc2e4 \ud568\uc218\ub294 \ubaa8\ub378\uc758 \uc608\uce21\uac12\uacfc \uc2e4\uc81c\uac12 \uac04\uc758 \ucc28\uc774\ub97c \uce21\uc815\ud558\ub294 \ud568\uc218\uc785\ub2c8\ub2e4. \uc190\uc2e4 \ud568\uc218\uc758 \uac12\uc774 \ub0ae\uc744\uc218\ub85d \ubaa8\ub378\uc758 \uc131\ub2a5\uc774 \uc88b\ub2e4\ub294 \uac83\uc744 \uc758\ubbf8\ud569\ub2c8\ub2e4.<\/p>\n<\/li>\n \uc5ed\uc804\ud30c\ub294 \uc2e0\uacbd\ub9dd\uc5d0\uc11c \uc190\uc2e4 \ud568\uc218\ub97c \ucd5c\uc18c\ud654\ud558\uae30 \uc704\ud574 \uac00\uc911\uce58\ub97c \uc5c5\ub370\uc774\ud2b8\ud558\ub294 \uc54c\uace0\ub9ac\uc998\uc785\ub2c8\ub2e4. \uae30\uc6b8\uae30 \ud558\uac15\ubc95\uc744 \uc0ac\uc6a9\ud558\uc5ec \uac00\uc911\uce58\ub97c \uc870\uc815\ud569\ub2c8\ub2e4.<\/p>\n<\/li>\n \uacfc\uc801\ud569\uc740 \ubaa8\ub378\uc774 \ud559\uc2b5 \ub370\uc774\ud130\uc5d0 \ub108\ubb34 \uc798 \ub9de\uc544 \uc0c8\ub85c\uc6b4 \ub370\uc774\ud130\uc5d0 \ub300\ud55c \uc77c\ubc18\ud654 \uc131\ub2a5\uc774 \ub5a8\uc5b4\uc9c0\ub294 \ud604\uc0c1\uc785\ub2c8\ub2e4. \uc774\ub97c \ubc29\uc9c0\ud558\uae30 \uc704\ud574 \uc815\uaddc\ud654 \uae30\ubc95\uc744 \uc0ac\uc6a9\ud569\ub2c8\ub2e4.<\/p>\n<\/li>\n \ud558\uc774\ud37c\ud30c\ub77c\ubbf8\ud130\ub294 \ubaa8\ub378 \ud559\uc2b5 \uacfc\uc815\uc5d0\uc11c \uc124\uc815\ud574\uc57c \ud558\ub294 \ud30c\ub77c\ubbf8\ud130\ub85c, \uc608\ub97c \ub4e4\uc5b4 \ud559\uc2b5\ub960, \ubc30\uce58 \ud06c\uae30 \ub4f1\uc774 \uc788\uc2b5\ub2c8\ub2e4. \ud558\uc774\ud37c\ud30c\ub77c\ubbf8\ud130\uc758 \uc120\ud0dd\uc5d0 \ub530\ub77c \ubaa8\ub378\uc758 \uc131\ub2a5\uc774 \ud06c\uac8c \ub2ec\ub77c\uc9c8 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n<\/li>\n<\/ul>\n<\/section>\n \ub2e4\uc74c\uc740 \ud30c\uc774\ud1a0\uce58\ub97c \uc0ac\uc6a9\ud558\uc5ec \uae30\ubcf8\uc801\uc778 \uc778\uacf5 \uc2e0\uacbd\ub9dd\uc744 \uad6c\uc131\ud558\uace0 MNIST \uc22b\uc790 \uc778\uc2dd \ub370\uc774\ud130\ub97c \ud559\uc2b5\ud558\ub294 \ucf54\ub4dc\uc785\ub2c8\ub2e4.<\/p>\n \uc774 \ucf54\ub4dc\ub294 MNIST \ub370\uc774\ud130\uc14b\uc744 \uc0ac\uc6a9\ud558\uc5ec \uc190\uae00\uc528 \uc22b\uc790\ub97c \ubd84\ub958\ud558\ub294 \uc544\uc8fc \uac04\ub2e8\ud55c \uc778\uacf5 \uc2e0\uacbd\ub9dd \ubaa8\ub378\uc744 \ud559\uc2b5\ud558\ub294 \uc608\uc81c\uc785\ub2c8\ub2e4. neural network \uad6c\uc131 \uc804\ubc18\uc5d0 \uac78\uccd0 \uc190\uc2e4 \uacc4\uc0b0, \uc5ed\uc804\ud30c \ubc0f \uac00\uc911\uce58 \uc5c5\ub370\uc774\ud2b8 \uacfc\uc815\uc774 \uc774\ub8e8\uc5b4\uc9d1\ub2c8\ub2e4.<\/p>\n<\/section>\n \ub525\ub7ec\ub2dd\uc740 \ucd5c\uadfc \uba87 \ub144 \ub3d9\uc548 \uae09\uc18d\ud788 \ubc1c\uc804\ud558\uace0 \uc788\uc73c\uba70, \ud2b9\ud788 \uc790\uc5f0\uc5b4 \ucc98\ub9ac\ub098 \uc774\ubbf8\uc9c0 \ucc98\ub9ac \ubd84\uc57c\uc5d0\uc11c \uad04\ubaa9\ud560 \ub9cc\ud55c \uc131\uacfc\ub97c \ubcf4\uc5ec\uc8fc\uace0 \uc788\uc2b5\ub2c8\ub2e4. Transformer \ubaa8\ub378, GAN(Generative Adversarial Networks), \uac15\ud654\ud559\uc2b5(Deep Reinforcement Learning) \ub4f1\uc758 \uae30\uc220\ub4e4\uc740 \ubbf8\ub798\uc758 \ucca8\ub2e8 \uae30\uc220\ub85c \uc790\ub9ac \uc7a1\uace0 \uc788\uc73c\uba70, \ub2e4\uc591\ud55c \uc0b0\uc5c5\uc5d0 \uc751\uc6a9\ub420 \uc218 \uc788\uc2b5\ub2c8\ub2e4. \ub610\ud55c, \ud6a8\uc728\uc801\uc778 \uc790\uc6d0 \uc0ac\uc6a9, \ud658\uacbd \uce5c\ud654\uc801\uc778 \ud559\uc2b5 \ubc0f \ubaa8\ub378 \uacbd\ub7c9\ud654 \ub4f1\uc758 \uc5f0\uad6c\uac00 \ud65c\ubc1c\ud788 \uc9c4\ud589\ub418\uace0 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n<\/section>\n2. \ud30c\uc774\ud1a0\uce58(PyTorch)\ub780?<\/h2>\n
3. \ub525\ub7ec\ub2dd\uc758 \uc8fc\uc694 \uc6a9\uc5b4<\/h2>\n
\n
3.1 \uc778\uacf5 \uc2e0\uacbd\ub9dd (Artificial Neural Network, ANN)<\/h3>\n
3.2 \uc190\uc2e4 \ud568\uc218 (Loss Function)<\/h3>\n
3.3 \uc5ed\uc804\ud30c (Backpropagation)<\/h3>\n
3.4 \uacfc\uc801\ud569 (Overfitting)<\/h3>\n
3.5 \ud558\uc774\ud37c\ud30c\ub77c\ubbf8\ud130 (Hyperparameter)<\/h3>\n
4. \ud30c\uc774\ud1a0\uce58 \uc608\uc81c \ucf54\ub4dc<\/h2>\n
4.1 \uae30\ubcf8 \uc778\uacf5 \uc2e0\uacbd\ub9dd \uad6c\uc131\ud558\uae30<\/h3>\n
\nimport torch\nimport torch.nn as nn\nimport torch.optim as optim\nfrom torchvision import datasets, transforms\n\n# \ub370\uc774\ud130\uc14b \ub85c\ub4dc\ntransform = transforms.Compose([transforms.ToTensor()])\ntrain_dataset = datasets.MNIST(root='.\/data', train=True, download=True, transform=transform)\ntrain_loader = torch.utils.data.DataLoader(dataset=train_dataset, batch_size=64, shuffle=True)\n\n# \uc778\uacf5 \uc2e0\uacbd\ub9dd \ud074\ub798\uc2a4 \uc815\uc758\nclass SimpleNN(nn.Module):\n def __init__(self):\n super(SimpleNN, self).__init__()\n self.fc1 = nn.Linear(28 * 28, 128)\n self.fc2 = nn.Linear(128, 10)\n\n def forward(self, x):\n x = x.view(-1, 28 * 28) # 1D\ub85c \ubcc0\ud658\n x = torch.relu(self.fc1(x)) # \ud65c\uc131\ud654 \ud568\uc218\n x = self.fc2(x) # \ucd5c\uc885 \ucd9c\ub825\n return x\n\n# \ubaa8\ub378, \uc190\uc2e4 \ud568\uc218 \ubc0f \uc635\ud2f0\ub9c8\uc774\uc800 \uc815\uc758\nmodel = SimpleNN()\ncriterion = nn.CrossEntropyLoss()\noptimizer = optim.SGD(model.parameters(), lr=0.01)\n\n# \ud559\uc2b5\nfor epoch in range(5): # 5 epoch \ud559\uc2b5\n for images, labels in train_loader:\n optimizer.zero_grad() # \uae30\uc6b8\uae30 \ucd08\uae30\ud654\n outputs = model(images) # \ubaa8\ub378 \uc608\uce21\n loss = criterion(outputs, labels) # \uc190\uc2e4 \uacc4\uc0b0\n loss.backward() # \uc5ed\uc804\ud30c\n optimizer.step() # \uac00\uc911\uce58 \uc5c5\ub370\uc774\ud2b8\n print(f'Epoch {epoch+1}, Loss: {loss.item()}')\n <\/code><\/pre>\n5. \ub525\ub7ec\ub2dd\uc758 \ubc1c\uc804 \ubc29\ud5a5<\/h2>\n